Mathematical Reasoning on GSM8k Tiny
31.7E*SRecover-LoRA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=OpenHermes-10k, Training Samples=10k2026.06 | 31.7 | — | — | 55 | |
| Recover-LoRAModel=Qwen3-4B, Quantization=W2-GateUp, Training Data Source=Synthetic-10k, Training Samples=10k2026.06 | 27.8 | — | — | 48 | |
| Quantized Student (W2-GateUp)Model=Qwen3-4B, Quantization=W2-GateUp (gate/up at INT2)2026.06 | — | — | 1.5 | — | |
| Teacher (BF16)Model=Qwen3-4B, Precision=BF162026.06 | — | 56.3 | — | — |